What is the Problem to Which Fair Machine Learning is the Solution?

2017 Symposium

Solon Barocas

2017 Symposium

July 10, 2017
Experts Workshop

What is the Problem to Which Fair Machine Learning is the Solution?

A lightning talk from the Bias & Inclusion session of the 2017 Experts Workshop given by Solon Barocas (Cornell University).

In this talk, Solon looks at how the past year has seen a sudden rush of work on fairness in machine learning, with proliferating formal definitions of fairness and new techniques to build and field fair models. While scholars continue to debate the relative merits of these definitions and techniques, the field has been slow to reflect on the exact normative and policy problem to which fair machine learning can—or should—provide a solution. Solon will show that existing work on fair machine learning actually addresses a number of different problems, but that future progress will depend on a more coherent vision for the field and a better appreciation of the limits of each approach.

Topics
Bias & Inclusion